Facebook’s Algorithm is predicting COVID-19 Deterioration
How Facebook’s ML models are predicting COVID-19 Deterioration?
During the COVID-19 pandemic, rapid and accurate triage of patients at the emergency department is essential in decision-making. Researchers at Cornell University propose a data-driven approach for automating deterioration risk using a deep neural network that learns from chest X-ray images and a gradient accelerating model that learns from routine clinical variables. Its AI prognosis system, trained using data from 3,661 patients, achieve an area under the receiver operating characteristic curve of 0.786 while predicting deterioration within 96 hours.
Similarly, social media giant Facebook has published research conducted by its artificial intelligence (AI) unit on predicting COVID-19 deterioration in collaboration with NYU Langone Health’s Predictive Analytics Unit and Department of Radiology.
It says that it has developed three machine learning (ML) models that could help doctors predict how a patient’s condition may develop, and to help hospitals ensure they have sufficient resources to care for patients. The first ML model has been developed for predicting patient deterioration based on a single X-ray. The second model helps in predicting patient deterioration based on a single of X-rays. The last model predicts how much supplemental oxygen a patient might need based on a single X-ray.
Facebook claims that these predictions could help doctors avoid sending at-risks patients home soon and hospitals to predict demand for supplement oxygen and other limited resources better.
Based on reader studies, Facebook’s ML models that used sequences of X-ray images outperformed human experts at predicting intensive care unit needs and mortality predictions and overall adverse event predictions up to 96 hours. Being capable of predicting whether a patient will need oxygen resources would also be a first and could help hospitals as they decide how to allocate resources in the weeks and months to come.
Facebook claims that its research covers the full spectrum of topics related to AI, deriving knowledge from data: theory, algorithms, applications, software infrastructure and hardware infrastructure. Long-term objectives of understanding intelligence and creating intelligent machines are bold and ambitious, and the company knows that making significant progress towards artificial intelligence cannot be done in isolation. To do this, Facebook AI researchers actively engage with the research community through publications, open-source software, participation in technical conferences and workshops.
The researchers have built adaptive models and collaborated with experts to help the world better understand the virus and its spread. AI is an important tool to support public health experts across the world to save lives. Facebook AI is collaborating with academic researchers and other experts on a range of initiatives associated with COVID-19, and some of them have been observed over the past few months.
The company in October 2020 published AI-infused forecasts that predicted the spread of COVID-19 across the United States to understand the virus’s geographic progression better and help healthcare providers prepare for surges in their regions.
Apart from Facebook, similar AI research strategies are being used by numerous companies, including Israel-based Zebra Medical that uses AI and ML algorithms for the automated analysis of radiological scans like X-rays, CT scans, and MRI.